Adsl Speedtests Which One Is Right?

[jfchandler]

Another case of misguided thinking...

Just because some big company doesn't have it's own internet pipeline into Thailand (and I clearly agree on that point) doesn't mean that every time someone accesses their data it's coming from the company's home location in the U.S.

Awhile back, as I recall, you were boasting about getting good speeds on an Apple software download, presuming that it was coming from their Bay Area California HQ. And I pointed out that outfits like Google, MS and Apple clearly keep their public data on servers in various international locations -- not just in their headquarters location -- to both speed up file access to users and I assume provide some redundancy....

They do also, I'm sure, have very high and fast server capacity... But in all likelihood, the file was arriving quickly because it probably was physically coming from someplace a whole lot closer to Thailand, or maybe even inside Thailand, than Apple's HQ.

And then there's the issue of the Thai ISPs themselves caching certain Internet content on their own, in order to make it available here locally more quickly, and I presume not overly taxing their international bandwidth with repeated requests for the same content.

[RedCardinal]

So you do realise that all the major Internet players have direct pipes into Thailand?

em I don't think so dude. Microsoft/Google/Apple/Sun/Cisco do not have their own "Direct Pipes" into Thailand. However do please prove me wrong.

Yet another conspiracy theory.

You are getting the decent speeds from such providers simply because they have the bandwidth available. Do a tracecert it is certainly not their own "direct pipe into Thailand"

Do some research on peering. You'll find companies like Google, MS peer directly with networks in various countries. Peering means they don't share the same bandwidth/network as regular traffic. Google has peering agreements in about 100 countries, of which I'm pretty sure Thailand is one. I'll be honest and say I'm not so sure about MS, but if anything I imagine their needs are similar to Google, if not greater due to the bandwidth consumed by updates to their OS installs. The other thing worth noting is that most major players have distributed DC networks around the world.

Even if your tracert shows packets coming from US it doesn't mean that you're using the same bandwidth as everyone else. But feel free to prove me wrong...

[negreanu]

They do also, I'm sure, have very high and fast server capacity... But in all likelihood, the file was arriving quickly because it probably was physically coming from someplace a whole lot closer to Thailand, or maybe even inside Thailand, than Apple's HQ.

Apple Itunes US Content is only Hosted on their Server Farms in the US. Full stop. They do not MIRROR US CONTENT around the world - You know how big their server farm is?? You cannot just mirror a server farm of that capacity. There current data centre is 107,000sqft!

Apples New Data Centre:

Your just making a further fool of yourself. Here are the addresses of the two ITUNES STORE servers. I am sure you can do a traceroute......

phobos.apple.com

phobos.apple.com.edgesuite.net

Which is exactly where my US content from itunes to BKK that maximises my bandwidth comes from.

Edited July 29, 2010 by negreanu

[RedCardinal]

Just because some big company doesn't have it's own internet pipeline into Thailand (and I clearly agree on that point)...

So think about that for a minute. All the Thai folk on the regular Thai Internet packages that have crap intl access would be getting regular timeouts accessing Google.co.th. Google's nearest DC is Singapore, so if Google doesn't have direct peering into Thailand how is it all the Thai internet users on ISP's "internal" packages can access Google.co.th so quickly?

(Yes, I do know about direct peering between Thailand to Singapore, but the example would be the same if Google's nearest DC was in HK)

[jfchandler]

Thanks for the other comments on this... I'm not an expert on this, but suffice to say, it still leads me to believe getting a file quickly from Apple isn't the same as having a fast international connection from a Thai ISP.

Edited July 29, 2010 by jfchandler

[negreanu]

At the end of the day putting aside the infrastructure of the net. The Question is can you maximise 16mb+ connection in Thailand using everyday uses of the internet. The answer is YES to the following applications of the internet (The ones I have tried and can verify):

1. Downloads from servers that have the available bandwidth. (No matter how it gets here)

2. Torrents

3. Itunes Store / Amazon Download store.

4. Usenet server downloads

5. Online gaming not bandwidth related but yes u can to the us with 230ms pings.

6. Microsoft/Google/Yahoo/Sun/Adobe downloads

7. 720p You Tube Streaming Video.

8. Skype/ichat video to video chat/

9. Game Downloads for example Blizzards products.

10. Netflix and Hulu (Quality of your VPN depending of course)

I am sure there are some other applications of the internet that require high bandwidth but from a normal consumer point of view if you can do all that and maximise your bandwidth here in Thailand. What else do you currently need that high bandwidth for..?

Edited July 29, 2010 by negreanu

[RedCardinal]

Apple Itunes US Content is only Hosted on their Server Farms in the US. Full stop. They do not MIRROR US CONTENT around the world - You know how big their server farm is?? You cannot just mirror a server farm of that capacity. There current data centre is 107,000sqft!

Your example of "Apple US Content" might suit your argument, but it's plain daft to assert that large DCs cannot be mirrored globally. Just because Apple don't do it doesn't mean it's not possible. Do you know how many DCs Google has globally? Amazon? MS? In some cases it may be difficult due to the types of data being stored, but then you'll often find large corps using CDNs. Apple may not use global CDNs (although they are certainly using Akamai as edgesuite.net is an akamai domain), but that doesn't preclude other companies from doing so.

[RedCardinal]

The Question is can you maximise 16mb+ connection in Thailand using everyday uses of the internet.

OK, serious question - are you talking about 16Mb from someone like True who claim full intl bandwidth, or 16Mb from someone like 3bb who offer "Thai internet"?

I'm genuinely interested.

[negreanu]

The Question is can you maximise 16mb+ connection in Thailand using everyday uses of the internet.

OK, serious question - are you talking about 16Mb from someone like True who claim full intl bandwidth, or 16Mb from someone like 3bb who offer "Thai internet"?

I'm genuinely interested.

I am saying that on True 16mb Premium product i can achieve more or less that bandwidth from downloading from the US. much to the disbelief of a couple of the users on here. (Even after posting screenshots of downloads in progress of Itunes downloading, Giganews Downloading etc)

So if we have agreed that Apple are not mirroring their US Content Mr Chandler and I can maximise my 16mb True connection from downloading from their US farm can we then say it is possible to download from the US at 16mb here in BKK?

it still leads me to believe getting a file quickly from Apple isn't the same as having a fast international connection from a Thai ISP.

So if none of the above applications satisfy your criteria of having a fast international connection from a Thai ISP. Could you please enlighten me exactly what your definition of fast international connection from a Thai ISP is?

Edited July 29, 2010 by negreanu

[jfchandler]

OK...now that you've stopped throwing insults, we can have a serious discussion...

I'm assuming you're talking about some kind of multi-threaded downloading to accomplish the higher bandwidth rates...

I use Hulu and Netflix, but I don't know any way to increase their throughput beyond their regular single incoming streams...

Since you mentioned them above, what exactly are you doing to accomplish that... I'd like to know, and I'm sure others would as well...if it's real...

[negreanu]

OK...now that you've stopped throwing insults, we can have a serious discussion...

I'm assuming you're talking about some kind of multi-threaded downloading to accomplish the higher bandwidth rates...

I use Hulu and Netflix, but I don't know any way to increase their throughput beyond their regular single incoming streams...

Since you mentioned them above, what exactly are you doing to accomplish that... I'd like to know, and I'm sure others would as well...if it's real...

Hulu and Netflix streaming depends on your VPN of course. Obviously no matter how fast your connection here it will be limited by latency, your VPN etc. So for the best results on Hulu and netflix I would try VyprVPN or StrongVPN. Remember both of those products are not designed to be used outside the US but can using the above method but you cannot expect perfection using a VPN for streaming services.

However having said that VyprVPN and StrongVPN offer as about the best options you will have. Secondly, (If i remember correctly - i would have to do a search) a 4MB connection in US will stream Netflix and Hulu at their maximum quality.

The issue of Hulu and Netflix here in Thailand is not so much a Bandwidth thing (Your 8mb will stream as well as a 16mb for these applications) its down to latency etc.

I was not insulting but you are being very narrow-minded because you cannot get the bandwidth you need or pay for that anyone who says they have it are obviously lying!

[jfchandler]

And I would counter that your posts were overly simplistic, making it sound that anyone who has an 8 mb or 16 mb True internet connection can just log onto the internet here in Thailand and expect to get all their connections to the U.S. at those speeds....without having to take other measures...

Anyone using Netflix or Hulu here pretty much needs a VPN or at least a proxy in order to obtain a U.S. IP... And I'm familiar with the VPNs you mention.... But I haven't seen you post what kind of stream speeds you get using whatever one(s) you're using when accessing Netflix or Hulu.... And there is no multi-threading, that I'm aware of, when using those services... But please correct me if I'm wrong about that.

I'm watching MSNBC now via a similar method online, and monitoring the stream speeds via Bitmeter... The numbers are bouncing around between 300 and 1300 kbs, and most of the time falling in the 400 to 800 kbits range...this being at noon time on a weekday... Wouldn't get that kind of performance here in the evenings... That's .4 to .8 Mbits (less than 1 Mbit) on a True 8 Mbit plan...

I'd challenge you to monitor the speed of the stream you're getting from Netflix or Hulu in kbit/mbit terms, and post the real results here...

And of course, you don't seem willing to discuss the details of whatever multi thread file downloading approach you're using for other things of things....

All I'm saying is if someone wants to make claims here, they should at least post the proof and method so others can follow that example.

[negreanu]

Only multi threaded downloading I am using is for usenet and torrents. The rest is not multi threaded. You cannot obviously multi thread streaming content

You cannot expect to maximise your bandwidth with the US ONLY Hulu and Netflix through a VPN to the otherside of the world - If Hulu an Netflix are the only reason you are complaining about True premium products then that is just wrong.

In a nutshell you are blaming True for bad streaming for a US ONLY service through a Third Party VPN on the other side of the world and you are wondering why performance is poor. You cannot seriously be blaming True for this bad performance.

Hulu and Netflix were never designed to be used outside the US nevermind through a VPN in the US then relayed to the otherside of the world.

Example I can stream Hulu through a VPN when I am in Korea (80mb Connection). It is just as bad/good as my True Premium connection streaming Hulu in Thailand! You are trying to do something that the provider both Hulu an the VPN provider and the ISP were never designed to do.

Look to oversimplify again: Expecting the normal performance through a VPN on the other side of the world streaming data is never going to happen no matter how fast your connection is out here.

The only advise i could give you is just let the buffer fill on Hulu before you click play. That is as good as it will get.

Edited July 29, 2010 by negreanu

[Supernova]

Supernova, I'm not seeing any big variation between the results from the Java and Flash versions of DSLReports speed test, though I gather there can be particularly with higher bandwidth accounts...

The variation isn't that big, but it's there. Note: Since I'm on a very long and old TOT line, I set a higher target SNRM during work hours to maintain connection stability which lowers sync speed. Here are some test results...

Line Rate (down/up Kbps): 1856/512

dslreports - Los Angeles

Speedtest.net - Los Angeles

#1

#2

Speakeasy.net - Los Angeles

Download Speed: 1582 kbps (197.8 KB/sec transfer rate)

Upload Speed: 367 kbps (45.9 KB/sec transfer rate)

Edited July 29, 2010 by Supernova

[jfchandler]

Once again, you're wrongly stating what I've consistently said here...

I'm not blaming or criticizing True for anything. In fact, I specifically posted earlier that I'm a reasonably satisfied True DSL customer for what I pay and what I get...

What I have consistently said is that having an 8 mbit or 16 mbit plan/connection here in Thailand with True will usually get one only a portion of that speed, and particularly during the primetime night hours a small portion, for regular U.S. internet connections.... In my experience, typically 400-600 kbits.

Just a few posts earlier, you were boasting that you can get full use of your 16 mbit True account for your list of 10 uses, and you included Netflix and Hulu on that list... Now suddenly in the post just above here, you're coming up with all kinds of reasons why a Netflix or Hulu user here CAN'T expect to get full/good speeds...

When you make up your mind about which story you're going to stick with, please let us know..

And, if things are so great for you, why don't you monitor and specifically post here either a) results from a reliable speedtest site outside Thailand for a U.S.-Thai connection (as I've done dozens of times here) and/or evidence of what kind of stream speeds you get when using Netflix or Hulu. Only then will your claims begin to have any credibility...

You're very long on talking, and very short on providing any clear proof or evidence.

Only multi threaded downloading I am using is for usenet and torrents. The rest is not multi threaded. You cannot obviously multi thread streaming content

You cannot expect to maximise your bandwidth with the US ONLY Hulu and Netflix through a VPN to the otherside of the world - If Hulu an Netflix are the only reason you are complaining about True premium products then that is just wrong.

In a nutshell you are blaming True for bad streaming for a US ONLY service through a Third Party VPN on the other side of the world and you are wondering why performance is poor. You cannot seriously be blaming True for this bad performance.

At the end of the day putting aside the infrastructure of the net. The Question is can you maximise 16mb+ connection in Thailand using everyday uses of the internet. The answer is YES to the following applications of the internet (The ones I have tried and can verify):

1. Downloads from servers that have the available bandwidth. (No matter how it gets here)

2. Torrents

3. Itunes Store / Amazon Download store.

4. Usenet server downloads

5. Online gaming not bandwidth related but yes u can to the us with 230ms pings.

6. Microsoft/Google/Yahoo/Sun/Adobe downloads

7. 720p You Tube Streaming Video.

8. Skype/ichat video to video chat/

9. Game Downloads for example Blizzards products.

10. Netflix and Hulu (Quality of your VPN depending of course)[/quote]

Edited July 29, 2010 by jfchandler

[jfchandler]

Once again... NO!!!

Apart from anything to do with Netflix or Hulu, which is its own separate issue and discussion, I'm more broadly pointing out that my 8 mbit True connection typically will only rate 400-600 Kbits per second when measured by any range of independent speed test sites that measure the connection between BKK and the U.S.

I've repeatedly posted clear evidence of that from multiple and different well-regarded speed test sites outside of Thailand. I'm still waiting for you to show anything comparable for your connection... waiting....waiting... waiting....

In a nutshell you are blaming True for bad streaming for a US ONLY service through a Third Party VPN on the other side of the world and you are wondering why performance is poor. You cannot seriously be blaming True for this bad performance.

[negreanu]

ARGH!!!! you dont get it. Look at my screenshots posted previously what more do you want!!! You can see actual download rates of actual internet use! Its far more reliable than your 2nd rate speedtest apps.

I give up. I really do. Thats it ok you win my internet never gives me 16mb connection to the US. And anything that says it does is lying including downloads and speedtests. Enjoy your internet.

Edited July 29, 2010 by negreanu

[jfchandler]

OK...so you don't believe in speed test sites now.... Every time, a different story and a different excuse...

No monitoring of your stream speeds... no posting of any real results...once again.

Interestingly, right now using Bitmeter, I'm watching in real time the speed of Itunes podcast downloads to my PC... I don't know where the files are actually coming from...

The speeds fluctuate a lot, and drop off/ramp up at the beginning of each separate file. But the download speed numbers seem to be generally bouncing around between 2000 and 4000 kbits per second, with valleys and peaks between 1000 and 6000 Kbits...which is pretty good... Wonder how those file deliveries are being handled...

As others have suggested above, I suspect Apple is doing something to speed the delivery of its files to locations around the world, rather than simply trying to stream them from the U.S. using normal Internet channels...

Edited July 29, 2010 by jfchandler

[hml367]

EDIT - Removed... definitely not worth getting into the fray.

MSPain

Edited July 29, 2010 by hml367

[negreanu]

Ok. When I get back to BKK from abroad in a few days. If I post SCREENSHOT showing the following and in the same shot showing istats menu download meter will you believe that I can get 16mb from USA:

1. Torrent Download.

2. Apple itunes downloads.

3. Microsoft Download

4. Giganews West Coast Download

5. Speedtest.net result showing correct and valid PING for west coast download

6. Blizzard Game Download from West Coast Servers

Now the question is if they all show me maxing out my bandwidth with physical downloads. Then I run your DSLreports Speedtest which shows I am not. Are you seriously going to believe that over the above physical file downloads? I am not really sure how you could disregard all the above over a speedtest site.

The reason your speed is fluctuating with itunes downloads is the files are too small. Queue up say 4 300mb files and your speed will stabilise at your bandwidth limit. This is a similar problem with your DSLreports speedtest in that the file size for the test is too small for Thailand in that due to latency and the amount of links in the chain by the time a connection here in Thailand accelerates to capacity the speedtest is already over due small file size.

You can actually see this happening in Speedtest.net also in that the speed ramps up as the test continues towards the middle-end.

Edited July 29, 2010 by negreanu

[hml367]

Is there anyone else getting verifiable download speeds at something near what 'negreanu' is saying he gets?

MSPain

Ok. When I get back to BKK from abroad in a few days. If I post SCREENSHOT showing the following and in the same shot showing istats menu download meter will you believe that I can get 16mb from USA:

1. Torrent Download.

2. Apple itunes downloads.

3. Microsoft Download

4. Giganews West Coast Download

5. Speedtest.net result showing correct and valid PING for west coast download

6. Blizzard Game Download from West Coast Servers

Now the question is if they all show me maxing out my bandwidth with physical downloads. Then I run your DSLreports Speedtest which shows I am not. Are you seriously going to believe that over the above physical file downloads? I am not really sure how you could disregard all the above over a speedtest site.

The reason your speed is fluctuating with itunes downloads is the files are too small. Queue up say 4 300mb files and your speed will stabilise at your bandwidth limit. This is a similar problem with your DSLreports speedtest in that the file size for the test is too small for Thailand in that due to latency and the amount of links in the chain by the time a connection here in Thailand accelerates to capacity the speedtest is already over due small file size.

You can actually see this happening in Speedtest.net also in that the speed ramps up as the test continues towards the middle-end.

[jfchandler]

I'd be very and genuinely interested in whatever results you are willing to post, provided that you and/or they are clear in what they represent, and how they're being obtained...

I guess part of this discussion comes down to what do we mean when we talk about the speed of an Internet connection... In that, it may be one rate for one kind of thing, and a different rate for a different kind of thing, and so on and so on, as this thread has demonstrated... Generally, I think most people take the results of legitimate speed tests as the indicator of the real speed of their particular connection. But clearly in some cases as you've pointed out, that's not indicative of every use. So in the end, it comes down to what kind of speed a person can get for the particular things they want to do on the Internet... And those things may be different for different people.

I've clearly stated and posted many times here what my speed test results are connecting to U.S. servers, using a variety of speed test sites not gamed by True. Typically 400 to 600 kbits per sec in the evening hours in BKK, and double or triple that in the morning hours here. So that's about a half to 2 Mbits per sec. to the U.S. on an 8 Mbit plan.

The results from the Apple Itunes podcasts downloads this morning were interesting...in that they were running at a 2 - 4 Mbits speed as measured in real time by my Bitmeter program while running thru a VPN... and those rates were measured with about a half dozen podcast files lined up waiting in queue... regarding your earlier comment above. That's quite a bit faster than other generic file downloads I get from non-big time sites... Why is that? Because the other sites have slower servers or Apple has faster ones... Or because Apple is caching or peering their content somewhere other than in the S.F. Bay area.. I don't know... But in any case, that result is up to 4 Mbits on an 8 Mbit plan.

Re Netflix and Hulu, if those are tried anytime during the primetime hours here, they are having to rebuffer a lot... After midnight or so, they run fine. I haven't used Bitmeter to measure what the actual incoming stream speeds are, both in the primetime and non-primetime hours, but I will...and I'll post the results here...

But I guess the bottom line is, for all of those applications, the achieved speeds aren't getting anywhere near my True plan's rated speed of 8 Mbits, although that admittedly is an inside Thailand rating...and they don't claim that's what customers will be internationally.

I must say, though, your latest list of SIX, down from TEN earlier, is interesting... in that sources like Apple and Microsoft aren't likely to be indicative of general purpose Internet traffic and knowing the true location source for the file may be difficult. Torrents, of course, can involve multi-threading... But nonetheless, I'll be interested to see what evidence you have to offer.

[negreanu]

in that sources like Apple and Microsoft aren't likely to be indicative of general purpose Internet traffic

So in 2010 what is indicative of general purpose internet traffic if you exclude the 6 unique applications I gave as an example of high bandwidth usage?

Edited July 29, 2010 by negreanu

[jfchandler]

You really need to either actually read, or stop mischaracterizing, what I actually say...

I never said anything about suggesting to exclude all six of the downloading sources you mentioned above as being legitimate test points...

I did say, in my own words, that "sources like Apple and Microsoft aren't likely to be indicative of general purpose Internet traffic and knowing the true location source for the file may be difficult."

In other words, not many of the content sources anyone's likely to encounter on the Internet have as much computing/server power behind them as those two particular companies do... And, as others raised earlier in this thread, there's the issue of where their content is actually coming from...somewhere in the USA, or cached or peered a whole lot closer to Thailand somewhere...

I don't know the answer to that question. I'm just pointing out, to be objective, one shouldn't blindly assume, without some evidence supporting the fact, that content from places like Microsoft and Apple that's requested inside Thailand is actually coming from the U.S.A. Maybe it is, but if you want to use them as test cases for international bandwidth, you ought to show that that's in fact the case.

Conversely, as an example, when I do a speedtest with DSL Extreme based in So. Cal... I KNOW their servers are located there and only there, because that's the only location they serve and operate as a company, and their speed test clearly indicates you're connecting to a server located in XXX California (or the West Coast, as the case may be)...

If you want to show them as evidence, I certainly hope you'll document clearly where the content is coming from, not just make a blind assumption. That's only being fair, accurate and objective.

[Supernova]

AFAIK, big corporations have always used CDNs to distribute content as far back as the mid-1990s. Not to mention ISPs caching content for more efficient and faster delivery. I don't get what all this bickering is about. Web-based speed tests aren't always accurate per se. To measure speed with a high degree of accuracy, use FTP. I'm surprised no one has mentioned this yet. FTP may be "old school", but it's the most reliable form of testing IMO.

CDN network test:

http://cachefly.cachefly.net/100mb.test

Note - this should route to the server closest to you...

FTP N. America:

ftp://ftp1.optonline.net/

ftp://ftp.telus.net/pub/size-test/

FTP W. Europe:

ftp://ftp.ovh.net/test.bin

ftp://ftp.euronet.nl/pub/test/

ftp://ftp.ukc.mirrorservice.org/sites/

For best results, use an FTP client to conduct FTP tests.

Edited July 29, 2010 by Supernova

[hml367]

If you run a test on an ftp server, wouldn't that give you ftp speeds and not http speeds?

MSPain

AFAIK, big corporations have always used CDNs to distribute content as far back as the mid-1990s. Not to mention ISPs caching content for more efficient and faster delivery. I don't get what all this bickering is about.

[Supernova]

^ The speed difference is negligible.

[jfchandler]

Supernova, thanks for that suggestion....

Although I sometimes wonder, what this forum OUGHT to be about is sharing information and knowledge, and helping everyone do better with whatever they've got... not name calling and bickering...

I didn't know off the top what a CDN is...but now I see...

Content delivery network From Wikipedia, the free encyclopedia A content delivery network or content distribution network (CDN) is a system of computers containing copies of data, placed at various points in a network so as to maximize bandwidth for access to the data from clients throughout the network. A client accesses a copy of the data near to the client, as opposed to all clients accessing the same central server, so as to avoid bottleneck near that server.

Content types include web objects, downloadable objects (media files, software, documents), applications, real time media streams, and other components of internet delivery (DNS, routes, and database queries).

Does anyone know if and/how worldwide operations like Microsoft and Apple use such systems in the online delivery of their content?? The answer to that, of course, would inform whether we can use their download speeds as any kind of realistic indicator of USA to Thailand bandwidth speeds.

By the way, unlike Apple or Microsoft, I'd argue it's pretty safe to assume that content providers such as Netflix and Hulu have built their system architecture for delivering online content pretty much only in the U.S., since they do not make their content available to IPs outside the U.S. and actually make efforts to ensure they only serve within the U.S. So there would seem little/no purpose for them to be making such delivery arrangements outside the U.S. Thus I'd take the speed of their content as a better indicator of actual international speeds.

[jfchandler]

Supernova, for whatever it's worth, just now at about 8:30 p.m. I downloaded a file - test16 at 16,384 Kb - from the ftp://ftp1.optonline.net/ site in North America you listed, and the download speed based on my browser's reading started at about 200 KBytes per second and then gradually declined to finish out at about 140 Kbytes per second, or, in other terms, started at about 1600 Kbits and finished at about 1120 Kbits... That was with my Firefox browser and nothing else downloading... I'm not sure I have a standalone FTP client installed on my PC anymore.. It's been years since I FTP'd anything anywhere.

I did the same with the 512 MB file, and it started much faster, around 300 Kbs, then gradually fell down to under 100 Kbs, before bouncing back to fluctuate mostly between 130 and 180 Kbs per second...

That is a pretty common pattern I've noticed in other downloads via True's Internet. Things seem to start out with a relatively fast burst, but then the speed falls off pretty quickly and never gets back to the original rate, but instead bumps along at a slower rate until finish.

AFAIK, big corporations have always used CDNs to distribute content as far back as the mid-1990s. Not to mention ISPs caching content for more efficient and faster delivery. I don't get what all this bickering is about. Web-based speed tests aren't always accurate per se. To measure speed with a high degree of accuracy, use FTP. I'm surprised no one has mentioned this yet. FTP may be "old school", but it's the most reliable form of testing IMO.

CDN network test:

http://cachefly.cach....net/100mb.test

Note - this should route to the server closest to you...

FTP N. America:

ftp://ftp1.optonline.net/

ftp://ftp.telus.net/pub/size-test/

FTP W. Europe:

ftp://ftp.ovh.net/test.bin

ftp://ftp.euronet.nl/pub/test/

ftp://ftp.ukc.mirrorservice.org/sites/

For best results, use an FTP client to conduct FTP tests.

Edited July 29, 2010 by jfchandler

[jfchandler]

And... 10 pm download of 16 MB file from download.cnet.com... as measured by Firefox download window...with nothing else arriving.

Started a bit quicker, then quickly fell into the 150 to 160 KBytes per sec range through completion of the file...

So that's 1200 to 1300 Kbits per sec... in primetime BKK time... or... about 1.2 to 1.3 Mbits per sec on True's 8 Mb premium plan.